geochemical mapping of new mexico, usa, using stream sediment data

19
ORIGINAL ARTICLE Geochemical mapping of New Mexico, USA, using stream sediment data Taisser Zumlot Philip Goodell Fares Howari Received: 23 September 2008 / Accepted: 14 November 2008 / Published online: 6 January 2009 Ó Springer-Verlag 2008 Abstract The spatial analysis of geochemical data has several environmental and geological applications. The present study investigated the regional distribution of Al, Ba, Ca, Ce, Co, Cr, Cu, Fe, K, La, Li, Mg, Mn, Na, Ni, Pb, Sc, Th, Ti, U, V, and Zn elements in stream sediment samples from New Mexico State. These elements were studied in order to integrate them with geological and environmental characteristics of the area. Data are used from 27,798 samples that were originally collected during the national uranium resource evaluation (NURE) Hydrogeochemical and stream sediment reconnaissance (HSSR) program in the 1970s. The original data are available as U.S. Geological Survey Open-File Report 97-492. The study used a variety of data processing and filtering techniques that included univariate, bivariate, factor analyses and spatial analyses to transform the data into a useable format. Principal component analysis and GIS techniques are applied to classify the elements and to identify geochemical signatures, either natural or anthro- pogenic. The study found that the distribution of the investigated elements is mainly controlled by the bed rock chemistry. For example, along the Rio Grande rift and Jemez lineament a strong association between Co, Cr, Cu, Fe, Ni, Sc, Ti, V and Zn was observed and indicates that elements distribution in the area controlled by the mafic factor. The rare earth elements (REE) factor which is consists of Ce, La and U, also has strong, localized, clusters in the felsic centers in New Mexico. Keywords Geochemical mapping Á Stream sediments Á Principal component analysis Á GIS Á New Mexico, USA Introduction Regional geochemical mapping is the study of the chem- istry of the surface of the earth and is fundamental scientific knowledge. It can be considered as an important tool in environmental studies, geologic mapping and min- eral exploration. Objectives of regional geochemical mapping include: (1) elucidation of fundamental earth systems processes, (2) exploration for new mineral resources using geochemical anomalies, (3) determination of geochemical background values, and (4) identification of natural or man made chemical contamination. The United Nations and the International Union of Geological Sciences are working toward a systematic global geochemical database and a world geochemical atlas (IUGS 2006; Darnely 1995). Table 1 gives a summary of several regional geochemical mapping studies. The Wolfson Geochemical Atlas of England and Wales developed in the 1960s by J. S. Webb at the Applied Geochemistry Research Center in England, consisted of over 50,000 stream sedi- ment samples that were obtained over an area of about 66,800 square miles (Howarth and Thornton 1983). These T. Zumlot Center for Environmental Resource Management (CERM), University of Texas at El Paso, El Paso, TX 79968, USA P. Goodell Department of Geological Sciences, University of Texas at El Paso, El Paso, TX 79968, USA F. Howari (&) Bureau of Economic Geology, Jackson School of Geosciences, The University of Texas at Austin, Austin, TX 78758, USA e-mail: [email protected] 123 Environ Geol (2009) 58:1479–1497 DOI 10.1007/s00254-008-1650-0

Upload: independent

Post on 15-May-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

ORIGINAL ARTICLE

Geochemical mapping of New Mexico, USA, using streamsediment data

Taisser Zumlot Æ Philip Goodell Æ Fares Howari

Received: 23 September 2008 / Accepted: 14 November 2008 / Published online: 6 January 2009

� Springer-Verlag 2008

Abstract The spatial analysis of geochemical data has

several environmental and geological applications. The

present study investigated the regional distribution of Al,

Ba, Ca, Ce, Co, Cr, Cu, Fe, K, La, Li, Mg, Mn, Na, Ni, Pb,

Sc, Th, Ti, U, V, and Zn elements in stream sediment

samples from New Mexico State. These elements were

studied in order to integrate them with geological and

environmental characteristics of the area. Data are used

from 27,798 samples that were originally collected during

the national uranium resource evaluation (NURE)

Hydrogeochemical and stream sediment reconnaissance

(HSSR) program in the 1970s. The original data are

available as U.S. Geological Survey Open-File Report

97-492. The study used a variety of data processing and

filtering techniques that included univariate, bivariate,

factor analyses and spatial analyses to transform the data

into a useable format. Principal component analysis and

GIS techniques are applied to classify the elements and to

identify geochemical signatures, either natural or anthro-

pogenic. The study found that the distribution of the

investigated elements is mainly controlled by the bed rock

chemistry. For example, along the Rio Grande rift and

Jemez lineament a strong association between Co, Cr, Cu,

Fe, Ni, Sc, Ti, V and Zn was observed and indicates that

elements distribution in the area controlled by the mafic

factor. The rare earth elements (REE) factor which is

consists of Ce, La and U, also has strong, localized, clusters

in the felsic centers in New Mexico.

Keywords Geochemical mapping � Stream sediments �Principal component analysis � GIS � New Mexico, USA

Introduction

Regional geochemical mapping is the study of the chem-

istry of the surface of the earth and is fundamental

scientific knowledge. It can be considered as an important

tool in environmental studies, geologic mapping and min-

eral exploration. Objectives of regional geochemical

mapping include: (1) elucidation of fundamental earth

systems processes, (2) exploration for new mineral

resources using geochemical anomalies, (3) determination

of geochemical background values, and (4) identification of

natural or man made chemical contamination. The United

Nations and the International Union of Geological Sciences

are working toward a systematic global geochemical

database and a world geochemical atlas (IUGS 2006;

Darnely 1995). Table 1 gives a summary of several

regional geochemical mapping studies. The Wolfson

Geochemical Atlas of England and Wales developed in the

1960s by J. S. Webb at the Applied Geochemistry Research

Center in England, consisted of over 50,000 stream sedi-

ment samples that were obtained over an area of about

66,800 square miles (Howarth and Thornton 1983). These

T. Zumlot

Center for Environmental Resource Management (CERM),

University of Texas at El Paso, El Paso, TX 79968, USA

P. Goodell

Department of Geological Sciences,

University of Texas at El Paso,

El Paso, TX 79968, USA

F. Howari (&)

Bureau of Economic Geology,

Jackson School of Geosciences,

The University of Texas at Austin,

Austin, TX 78758, USA

e-mail: [email protected]

123

Environ Geol (2009) 58:1479–1497

DOI 10.1007/s00254-008-1650-0

samples were analyzed for 35 different chemical elements.

The survey helped to understand certain environmental

issues such as diseases like molibdicosis (Thornton 1983)

in cattle, and others. Salminen and Tarvainen (1995) have

produced regional geochemical maps of Finland. The entire

country was covered at a reconnaissance scale using glacial

till, ground water, surface water and stream sediment as

sampling media. The study indicated the presence of gold

and multiple-sulphide ore deposits. Many other countries

have followed this trend, and Mexico has recently com-

pleted a countrywide survey.

In the USA the national uranium resource evaluation

(NURE) program sampled over 250,000 sites in the conti-

nental US and analyzed them for up to 40 constituents. The

hydrogeochemistry and stream sediment reconnaissance

program (HSSR) was robust; lithogeochemistry and

groundwater were less extensive. The US Geological Sur-

vey eventually took responsibility of the NURE data and

sample repositories, retested 10% of the samples, and pro-

ceeded to put this Archival Data on the web (http://pubs.

usgs.gov/of/1997/ofr-97-0492/[.). In 2004, the USGS, in

collaboration with other agencies, conducted the national

geochemical survey (NGS) to produce a body of geo-

chemical data for the United States based primarily on

stream sediments, analyzed using a consistent set of meth-

ods. In 2004, the NGS included data covering 80% of the

land area of the US, including samples in all 50 states

(USGS 2004).

Several researchers have successfully used NURE data

(e.g. Bolivar 1980; Grossman 1998; Ried 1993; Cocker

1999). For example, Ried (1993) employed the (NURE)

stream sediment data to prepare a geochemical atlas of

North Carolina, USA. The North Carolina NURE database

consists of 6,744 stream sediment samples, 5,778 ground-

water samples, and 295 stream water analyses. Cocker

(1999) conducted geochemical mapping in Georgia, USA,

using NURE data as a tool for geological and environ-

mental studies, and mineral exploration. Results indicated

that bedrock geology and mineralization are the most

important variables which influence the stream sediment

and stream water geochemistry. Anthropogenic sources

influence the geochemistry to a lesser and more localized

extent. Ludington et al. (2006) studied the regional distri-

bution of arsenic and 20 other element in stream sediment

samples in northern Nevada and southeastern Oregon.

They used 10,261 samples from the NURE program in

their study area. They analyzed data using point maps and

provided interpolation between data points to construct

high- resolution raster images, which were correlated with

geographic and geologic information using a geographic

information system (GIS).

Statement of the problem

Large geochemical databases for regional geochemical

mapping (RGM) are very useful. Initially, one may think

of a map for every chemical, however, this is not an ade-

quate and need extensive data analysis. Such analyses

include detailed univarient analysis to identify normal and

Table 1 Examples of regional geochemical studies

References Location No. of

chemical

species

No. of

samples

Sample types Sample

density

(Km2)

The Wolfson Geochemical Atlas

of England and Wales

England and Wales 35 50,000 Stream sediment

NURE USA 397,609 Stream sediment 1/100

NURE USA 335,547 Water 1/100

Shacklette and Boerngen (1984) USA 40 1,300 Soil

Riemann and Filzmoser (2000) 10 countries around

the Baltic Sea

41 Soil 1/2,500

Xie and Ren (1993) China 39

Ried (1993) North Carolina 12,522 Stream sediment and

water

Cocker (1999) Georgia, USA 19 8,248 Stream sediment

The National Geochemical

Survey (NGS) (2004)

USA 1/289

Robinson et al. (2004) New England states 25 8,360 Stream sediment and

water

Ludington et al. (2006) Northern Nevada and

Southeastern Oregon

21 10,261 Stream sediment

1480 Environ Geol (2009) 58:1479–1497

123

anomalous behavior, and the normal group should be tested

for normality in both arithmetic and log transformed for-

mulations. When performing this on a large geochemical

database, several questions need to be addressed such as:

are the data lognormal? Beyond what concentration values

lie the outliers? What geochemical associations are present

in the data, and where are they located? Due to the fact that

earth’s surface is subdivided into drainage basins and dif-

ferent geologic materials, a typical regional geochemical

mapping study produce many possibilities and combination

of controlling factors to the geochemical behavior of ele-

ments, and the question becomes that of prioritizing the

combinations. Toward this end, spatial analyses by GIS

tools are important. With this in mind, the present study

deals with the aforementioned research elements and

questions for the entire State of New Mexico. An objective

is a comprehensive statistical and spatial analysis, with

interpretations, cautions, and directions ahead. This will

lead to a better understanding of the surface geochemistry

of New Mexico. Another objective is the development and

demonstration of a template to transfer product to the sci-

entific audience. Due to the large number of figures, and

tables the present paper will present figures for few

elements and the rest of elements and associated statistical

and spatial products of this RGM research are on the

data repository website at https://webspace.utexas.edu/

howarifm/www/NURE/1nm.htm/.

Study area

New Mexico is the fifth largest state in the USA, with a total

area of 121,412 square miles (314457.079 sq.km) and lies

between latitudes 32 and 37� and longitudes 103 and 109�W.

New Mexico is characterized by the following physiographic

provinces (Pazzaglia and Hawley 2004): (1) Great Plains in

the eastern third of the state, (2) San Juan Basin occupies all

of northwestern New Mexico, and this is the southeastern

portion of the much larger Colorado Plateau, (3) the Rio

Grande Rift runs south to north across the state, being more

narrow to the north, (4) Southern Rocky Mountains straddle

the Rift in the north, (5) Mogollon Volcanic Plateau, in west

central NM which is covered by massive ash flow tuffs, and

(6) the Basin and Range of southwestern NM. Figure 1

shows the general location of these provinces.

Stable Precambrian craton of North America underlies

the Great Plains. The same stable craton also underlies the

Colorado Plateau; however the Plateau has been elevated

several thousand feet. The present geology of NM is dom-

inated by the Rio Grande Rift, a major, young, extensional

event of North America. The cratonic crust has been domed

upward and infused with heat, and subsequent collapse has

produced the Rift Valley. The Southern Rocky Mountains

are a portion uplifted by Rift processes. Erosion eastward

from the Southern Rocky Mountains and the Rift zone in

general provided for the deposition of sediments onto the

Fig. 1 Digital shaded relief

map of New Mexico with

physiographic provinces

Environ Geol (2009) 58:1479–1497 1481

123

stable craton constituting the Great Plains as a large alluvial

fan. The Mogollon Volcanic Plateau is a silicic large

igneous province (SLIP) where widespread, short duration,

magmatism was associated with the RG Rift. Progressive

stretching of the crust and collapse by listric faulting also

produced the basins in southwestern NM, which became

filled with sediments, forming the Basin and Range prov-

ince. Young mafic volcanism is expressed as extensive

basalt flows. The detailed geology of NM is complex; for

more information refer to Mack and Giles 2004.

For the present study, a simplified geologic map is given

in Fig. 2. Drainage basin boundaries in New Mexico are

given in Fig. 3. Stream sediments are the medium in this

study. Actually, only a low percentage of the samples are

from streams with water. Most samples are sieved from

material in dry streams called arroyos. In mountainous

regions stream sediments are closer to their bedrock source,

and bedrock distribution may vary over short distances.

Stream sediment chemistry may have greater variance. On

the Great Plains or in the basins the stream sediments tend

toward homogenization. Natural chemical leaching also

takes place during sediment transport.

Dataset

The relevant, edited, database for New Mexico consists of

22 chemical elements. Sample locations are given in

Fig. 4. Data presentation and discussion here will be lim-

ited to two chemicals, Mg and Ce, for reasons of space.

These include one major and one trace element, and serve

to illustrate the manner of data presentation. Study of all

chemicals is given on the web site (https://webspace.

utexas.edu/howarifm/www/NURE/1nm.htm/). The full

stream sediment data and analytic methods for the national

uranium resource evaluation (NURE) program are avail-

able as U.S. Geological Survey Open-File Report

97–492.\http://pubs.usgs.gov/of/1997/ofr-97-0492/[. The

samples were taken during 1975–1979. Chemical analyses

were performed at both the Los Alamos Scientific Labo-

ratory (LASL) (now LANL, Los Alamos National

Laboratory) and at Oak Ridge (formerly ORGDP, now

ORNL, Oak Ridge National Laboratory). The NURE pro-

gram was terminated prematurely, no synthesis or

interpretation was accomplished, and the data and samples

were almost forgotten.

NURE data are organized into data sets that are arranged

by quadrangles of 1 9 2 degrees in area. Geochemical data

used in this study are from the Albuquerque, Aztec,

Brownfield, Carlsbad, Clifton, Clovis, Dalhart, Douglas,

Durango, El Paso, Fort Sumner, Gallup, Hobbs, La Junta,

Las Cruces, Raton, Roswell, Saint Johns, Santa Fe, Ship-

rock, Silver City, Socorro, Tucumcari, and Tularosa

quadrangles. The statewide database consists of 27,798

stream sediment sites. The location of stream sediment

samples from the NURE Program are shown in Fig. 4.

Fig. 2 Generalized geological

map of New Mexico

1482 Environ Geol (2009) 58:1479–1497

123

Average sample density in New Mexico is one per 11 km2.

Several areas have enhanced sample density, as seen on

this figure. They are (1) Estancia Valley Pilot Survey where

results from 2,992 sediment and 505 water samples are

published (LASL 1977); (2) Grants Special Study where

results from 3,569 sediment and 167 water samples were

Fig. 3 Drainage basin

boundaries of New Mexico

Fig. 4 Locations of stream

sediment samples from NURE

program (1975–1979) of New

Mexico

Environ Geol (2009) 58:1479–1497 1483

123

published (LASL 1981a); and (3) San Andres and Oscura

Mountains Detailed Survey where results from 884 stream

sediments were published (LASL 1981b).

Methodology

The methodology of this study is straightforward. The data

of interest were located and downloaded. Quadrangle data

from within state boundaries were selected. Data from

multiple quadrangles were merged into master state data-

base. The dataset were checked for missing values and

sample concentrations below detection limits. Then as

statistical analysis and spatial analysis were carried out.

The statistical analysis included univariate analysis,

bivariate analysis and multivariate analysis. Univarient

analysis includes identification of outliers in raw data.

Outliers were removed, and the remaining data are labeled

as no outlier (NO). Logrithmic transformations of raw and

NO data are made. Four sets of data are then subjected to

plotting of cumulative frequency, boxplots, and histograms

and a summary of observations were made. Data were

tested for normality as will be explained in subsequent

sections. Bivarient analysis consists of the generation of a

table of correlation coefficients and pairwise correlation

coefficients. Scattergrams of some data have been made.

Multivarient analysis consisted of dendograms and princi-

ple component or factor analysis, including factor score

coefficients for every sample.

The NURE data are joined to the point coverage created

by the GIS from latitude and longitude data for each

sampling point. Geochemical mapping of the data were

produced using ArcGIS. Maps showing the actual numer-

ical concentration of the elements at their locations are not

produced in regional geochemical mapping studies, rather,

points or sample locations are color-coded according to the

concentration ranges of the element, with the highest range

shown in hot colors, and the lowest ranges shown in cold

colors. The geochemical data for stream sediments then

were interpolated in grid format to provide a graphical

visualization of the regional variation in geochemical val-

ues. The method of spatial interpolation is the inverse

distance weighted (IDW). The IDW techniques were rec-

ommended in studies comparing interpolation methods

(Robinson et al. 2004). This method is applied with 12

neighboring samples used for estimation of each grid point.

The power of one has been chosen to acquire some degree

of smoothing effect. The color-scheme of these maps was

similar to that used in the point maps. These maps were

useful for defining regional trends and local anomalies and

providing a quick visual check of the data processing.

Spatial analysis produces many different types of maps.

In the present study, the maps were developed for each

chemical and are labeled by the chemical symbol and are

numbered, where (1) is a point map of raw data, (2) is IDW

grid map of the raw data, (3) is geologic polygon extraction

from the raw data and the average for each polygon is color

scaled, (4) is drainage basin polygon extraction from the raw

data and the average for each polygon is color scaled, (5) is a

point map of the outliers and their values. ArcMap� 9.1 was

used to display the final maps. Finally, factor score coeffi-

cients generated from multivarient statistics are plotted on

maps. The raw data are stored in a dBASE file (dbf format),

and basic calculations are performed using Microsoft

Excel�. Most of the statistical calculations are accomplished

with SPSS� software (version 11.5) and JMPIN� (version

3.2.6). The geochemical maps are produced with Arc GIS�

software (version 9.1), and Arc View� (version 3.3).

Detection and removal of outlier

Evaluation of anomalies requires cumulative frequency,

Normal quantile, outliers box, quantile box, and histogram

diagrams, and these are carried out by the JMP, SPSS (e.g.

Sall and Lehman 1996), and ArcMap. The outlier box

diagram (Fig. 5) displays the characteristics of the empir-

ical distribution for single elements at a glance: location,

spread, skewness, tail lengths and outliers (Tukey 1977).

Fig. 5 Definition of the outlier

box diagram

1484 Environ Geol (2009) 58:1479–1497

123

The Box represents 50% of ordered data stretching

between the lower hinge and the upper hinge, which rep-

resent the lower and the upper quartile of the data

respectively. The vertical bar in this box indicates the

median, which by its position depicts the symmetry or

skewness of the data. The outlier box diagram is used here

to define the threshold for anomalous values symbolized by

the upper fence (cutoff) which is found by adding a step

(1.5 of the h-spread) to the upper hinge. Another cutoff, the

lower fence is found by subtracting a step to the lower

hinge. The lower and upper whiskers extend to the two

most extreme data values that are still inside the fences.

This tool has been applied with success to regional geo-

chemical mapping (Reimann et al. 2004; Bounessah and

Atkin 2003; Kurzl 1988). This approach found to be better

than the mean ?2 or 3 standard deviation method, where

the mean is affected by outlier data. The outliers are then

extracted from the data, and the results are retested.

Test for normality

Normality and log normality is to be tested for each chemical

for both the raw data and the data without outliers. The

Kolmogorov–Smirnov (K–S) test measures the degree to

which a given data set follows aspecific theoretical distri-

bution (such as normal, uniform, or Poisson). The statistical

test of K–S is based on the largest absolute difference

between the observed and the theoretical cumulative dis-

tribution functions. The K–S test assumes that the

parameters (e.g., mean and standard deviation) of the test

distribution are specified in advance, whereas the Lilliefors

correction for the K–S test is applied when means and

variances are not known and must be estimated from the

data. Results were obtained for K–S test values performed

by JMPIN� statistical software which utilized the Lillierfos

correction automatically called the KSL test (Sall and

Lehman 1996). The KSL test is applied by JMPIN� if n

[2,000. If the p-value reported is less than 0.05 (or some

other alpha), then the distribution is not normal. Therefore it

is useful to use the normal quantile plot to help assess the

lack of normality in the distribution (Sall and Lehman 1996).

Data set classification

Outlier removal is based on outlier box diagrams. Data are

divided into Raw versus No-Outliers. Furthermore, it is of

interest to test both the numerical values and their logarithms,

leading to Log-Raw data and Log-No-Outlier data. The

division of the data set is into five classifications, which are:

(1) Raw data: original data with values below the

detection limit are replaced by half of the minimum

of the datasets.

(2) No-Outliers data: raw data with outliers removed.

(3) Log- Raw data: the base 10 logarithm of raw data set.

(4) Log- No-Outliers data: the base 10 logarithm of no-

outliers data set.

(5) Outliers data: The data that pass the whiskers where

the lower and upper whiskers extend to the two most

extreme data values that are still inside the fences of

the box diagram.

All diagrams are presented for every chemical in the

web site: https://webspace.utexas.edu/howarifm/www/

NURE/1nm.htm/ and examples are shown in Figs. 6, 7,

8, 9, 10, 11.

Multivarient analyses

In this study these analyses included cluster analysis (CA)

and principal component analysis (PCA). Cluster analysis

seeks to identify homogeneous subgroups of cases in a

population and identify a set of groups which both mini-

mize within-group variation and maximize between-group

variation. One of the general approaches to cluster analysis

is hierarchical clustering. The product of this approach is

dendrograms, also called tree diagrams, show the relative

size of the proximity coefficients at which cases were

combined. Whereas PCA is a technique to take linear

combinations of the original variables such that the first

principal component has maximum variation; the second

principal component has the next most variation subject to

being orthogonal to the first, and so on. Each principal

component is calculated by taking a linear combination of

an eigenvector of the correlation matrix with a standardized

original variable. The eigen values show the variance of

each component. The set of n principal components has the

same total variation and structure as the original variables.

In the present study principal component analysis is per-

formed to reduce a large number of variables to a smaller

number, and for further investigation of the relationships

between the elements. The principal components (PCs)

with eigenvalues larger than 1 are extracted with the PC

loadings rotated for the maximum variance. A total of five

PCs are extracted, which account for 73.53% of the total

variance as will be discussed in the subsequent sections.

Results

The present work examines several chemical elements,

which are Al, Ba, Ca, Ce, Co, Cr, Cu, Fe, K, La, Li, Mg,

Mn, Na, Ni, Pb, Sc, Th, Ti, U, V, and Zn. This list is

arrived at by editing; incomplete data sets are eliminated,

and the detection limit problem were addressed. Concen-

trations shown as below the detection limit (DL) are

Environ Geol (2009) 58:1479–1497 1485

123

replaced by half of the minimum detection limit of the

datasets for each element. Forty chemical constituents are

reported in the database. However, the data must be posted,

because the NURE Program was terminated prematurely,

and not all samples were analyzed for all constituents.

Initially editing has been accomplished as a result of the

process of cleaning and re-cleaning data from missing or

inconsistent parameters in the dataset. The original NURE

files were reformatted into two consistent database struc-

tures: one for water samples and a second for sediment

samples, on a quadrangle-byquadrangle basis, from the

original NURE files. In this study the reported elements

Fig. 6 Histograms and normal

quantile plots for No-Outliers

data set of Ce (the observed

values (in ppm or %) are plotted

on the x-axis, and values for a

normal distribution are plotted

on the y-axis)

1486 Environ Geol (2009) 58:1479–1497

123

were selected because of their availability throughout the

entire state of New Mexico, and quality of the data, which

are Al, Ba, Ca, Ce, Co, Cr, Cu, Fe, K, La, Li, Mg, Mn, Na,

Nb, Ni, Pb, Sc, Sr, Th, Ti, U, V, and Zn. Concentrations

shown as below detection limit are replaced by half of the

minimum of the datasets.

An inherited complication in the dataset is that the

analytical chemistry for some chemicals was carried out by

different laboratories. Sample backlogs at Los Alamos

resulted in Oak Ridge laboratories taking over samples

from specific areas for specific chemicals. Each laboratory

consistently produced slightly different numbers, for

Fig. 7 Histograms and normal

quantile plots for no-outliers

data set of Mg

Environ Geol (2009) 58:1479–1497 1487

123

Fig. 8 The dendrogram (cluster

tree) for the studied elements

Fig. 9 Point, IDW, geologic

and hydrologic polygons maps

for cerium in New Mexico

1488 Environ Geol (2009) 58:1479–1497

123

several chemicals. The results appear on maps, where

usually a 1 9 2 degree quadrangle boundary or a smaller

boundary appears as a line separating areas of slightly

different shading. When this problem is recognized, the

results of one laboratory could be calibrated to another data

set by a simple algorithm; and this is one of the limitations

of the dataset. A reason for rejecting some data is too many

samples being below the minimum detection limit.

Univarient analysis

The filtered data matrix for the studied chemical elements

is present in the data repository website of this paper

https://webspace.utexas.edu/howarifm/www/NURE/1nm.

htm/. These data are here subjected to several types of

statistical analyses as presented in Tables 2, and 3. How-

ever, univarient results for Mg and Ce are given in Figs. 6

and 7. The proceed data repository noted earlier shows that

elements such as Ba, Ce, Cu, Fe, La, Mn, Ni, Pb, Th, U, V,

and Zn have high concentration values. These high con-

centrations can be influenced by anthropogenic processes.

However, inadequate detection limits are observed for 13

elements: Ag, Au, Bi, Cd, Cl, Hf, Mo, Sb, Se, Sn, Ta, Tb,

and W. This is defined as the 50th percentiles for all these

elements are below detection limits. The element Be does

not have good precision (due to lack of analytical resolu-

tion along with only one significant figure). For example,

5–50th percentiles of Be are all equal to 1 ppm. These 14

elements are regarded as having inadequate data quality,

and are not used in the statistical analyses.

One the other hand, the 25th percentiles of concentra-

tions for all the other elements were above the detection

limits thus they are regarded as of adequate quality for this

study. This study will investigate 24 elements that cover

the entire state of New Mexico: Al, Ba, Ca, Ce, Co, Cr,

Cu, Fe, K, La, Li, Mg, Mn, Na, Nb, Ni, Pb, Sc, Sr, Th, Ti,

U, V, and Zn. Many elements have values below or close

to the detection limits (e.g., Co, Nb, Ni, Pb, Sr, Th, and

Zn) resulting in the high frequencies for the lowest value

group, near that limit of detection. The frequency distri-

butions of most of the elements (except for Al) are

positively skewed and include some very high values.

Some extreme values appear to be separated from the

majority of the samples, and do not appear to be part of a

continuous distribution. These extreme values might be

regarded as evidence of mineralization or anthropogenic

processes.

Bivariant analysis

Bivariate analyses in this study are correlation coefficients,

pairwise correlations, correlation scatter diagrams, and

correlation frequencies. These analyses are carried out for

the logarithmic transformation of the No-Outlier data set

because it is closer to the normality condition which is

required for correlation analysis. Correlation frequencies

Fig. 10 Point, IDW, geologic

and hydrologic polygons maps

for magnesium in New Mexico

Environ Geol (2009) 58:1479–1497 1489

123

depend on a coefficient value greater than 0.29, and is the

number of other elements with which they correlate above

this number. Ca, Pb and Th have frequencies less than 5, K,

Mg, Na, and U have frequencies between 5 and 9, and all

other elements have frequencies greater than 10. Bivarient

analyses are not presented here, but are at the data

Fig. 11 Spatial distribution of

cerium and magnesium outliers

in New Mexico

1490 Environ Geol (2009) 58:1479–1497

123

Ta

ble

2D

escr

ipti

ve

stat

isti

cso

fth

era

wd

ata

(Al,

Ca,

Fe,

K,

Mg

,N

aar

ein

wei

gh

tp

erce

nt;

all

oth

ers

are

inp

pm

)

Ele

men

tn

Min

5%

10

%2

5%

Med

ian

75

%9

0%

95

%M

axM

ean

SD

Sk

ewn

ess

Ku

rto

sis

Al

25

,03

50

.02

52

.83

.33

4.1

45

5.0

45

.99

6.8

67

.31

51

8.2

85

.06

1.3

90

.03

0.5

1

Ba

25

,03

52

31

33

64

44

35

41

66

28

02

.49

16

34

57

05

85

.98

45

2.7

13

6.6

02

14

7.6

7

Ca

25

,03

20

.02

50

.37

0.4

90

.79

1.5

93

.14

5.7

37

.83

29

.97

2.4

92

.66

2.6

71

0.4

9

Ce

25

,03

33

.52

73

34

45

87

29

01

09

10

00

61

.70

30

.99

5.0

17

6.0

3

Co

24

,95

30

.35

0.3

53

.45

79

.21

31

6.8

71

.97

.72

5.0

82

.58

14

.60

Cr

25

,03

10

.51

31

82

43

24

56

17

88

15

38

.19

27

.34

5.4

46

9.5

9

Cu

24

,89

71

71

01

41

92

53

33

91

42

26

21

.94

95

.10

13

4.9

92

00

05

.49

Fe

25

,03

40

.04

0.9

01

.13

1.5

52

.07

2.7

63

.79

4.8

74

9.8

22

.40

1.6

86

.22

97

.07

K2

5,0

35

0.0

10

.86

1.0

11

.24

61

.51

1.8

2.0

76

2.2

49

7.3

04

1.5

30

.44

0.5

73

.50

La

25

,03

21

12

15

21

27

34

43

52

46

72

9.2

21

6.4

06

.55

10

0.8

6

Li

21

,13

40

.51

31

51

92

43

13

94

62

24

26

.32

11

.78

2.9

42

1.3

6

Mg

25

,03

50

.02

50

.20

40

.29

0.4

20

.63

60

.94

1.3

31

.68

7.6

70

.75

0.5

02

.19

9.6

6

Mn

25

,03

52

16

42

03

28

13

94

55

67

57

.49

10

29

,83

04

55

.53

34

4.5

22

8.2

22

16

4.0

7

Na

25

,03

50

.02

0.3

31

0.4

30

.61

0.8

31

.11

1.4

61

.69

12

4.3

40

.90

0.4

21

.12

2.3

7

Nb

24

,89

72

22

26

11

16

21

26

27

.73

7.9

14

.82

75

.68

Ni

24

,89

71

11

11

11

72

43

17

42

11

.91

12

.75

10

.07

45

5.0

4

Pb

24

,89

72

.52

.52

.59

14

21

28

33

96

91

17

.96

92

.54

83

.53

79

48

.41

Sc

25

,03

40

.05

2.6

34

67

.61

01

1.4

48

.86

.21

2.9

41

.66

7.3

6

Sr

25

,03

43

33

31

19

22

03

78

49

76

,06

71

55

.62

22

0.8

04

.99

59

.34

Th

25

,03

30

.55

0.5

53

58

11

14

17

.73

32

.59

.00

8.5

81

0.3

02

29

.29

Ti

24

,40

49

.51

45

4.2

51

77

82

28

0.2

52

93

0.5

38

99

51

98

.56

63

5.5

47

,68

03

38

2.6

32

15

3.7

75

.03

51

.06

U2

7,3

51

0.1

1.7

76

22

.42

.93

.55

4.7

08

6.1

04

44

5.1

3.3

84

.06

56

.48

54

25

.28

V2

5,0

35

12

63

24

25

57

21

01

13

31

,51

26

4.6

64

7.3

76

.42

89

.35

Zn

24

,88

61

.51

.51

.53

35

37

41

01

12

41

2,6

68

58

.89

14

4.9

26

1.5

34

59

2.5

6

Environ Geol (2009) 58:1479–1497 1491

123

Ta

ble

3D

escr

ipti

ve

stat

isti

cso

fth

en

o-o

utl

iers

dat

a(A

l,C

a,F

e,K

,M

g,

Na

are

inw

eig

ht

per

cen

t;al

lo

ther

sar

ein

pp

m)

Ele

men

tn

Min

5%

10

%2

5%

Med

ian

75

%9

0%

95

%M

axM

ean

SD

Sk

ewn

ess

Ku

rto

sis

Al

24

,82

11

.38

2.8

53

.36

94

.15

75

.04

5.9

86

.84

7.2

78

.74

5.0

71

.33

-0

.01

-0

.25

Ba

23

,97

01

16

31

83

65

44

15

35

64

77

62

82

79

89

54

9.2

61

52

.77

0.3

1-

0.1

3

Ca

23

,00

70

.05

0.3

80

.49

0.7

61

.45

2.6

84

.25

.16

6.6

61

.92

1.4

91

.18

0.6

9

Ce

23

,95

33

.52

73

34

45

77

08

39

21

13

57

.42

19

.61

0.2

3-

0.0

4

Co

21

,79

50

.73

.74

57

91

1.1

13

15

.47

.26

2.8

10

.65

0.0

8

Cr

23

,47

41

14

18

24

32

43

54

61

76

33

.93

14

.16

0.5

9-

0.0

1

Cu

23

,29

72

91

01

41

82

43

03

44

11

9.4

17

.60

0.5

6-

0.1

0

Fe

23

,57

00

.08

0.8

91

.11

1.5

12

.01

2.5

97

3.2

83

.73

4.5

82

.11

0.8

40

.54

0.0

3

K2

4,6

60

0.4

20

.87

1.0

12

1.2

51

.51

1.7

92

.05

42

.20

62

.62

91

.52

0.4

00

.11

-0

.25

La

23

,72

52

13

16

21

27

33

40

44

53

27

.28

9.4

00

.27

-0

.15

Li

20

,33

02

13

15

19

24

30

37

41

48

24

.82

8.2

60

.53

0.0

1

Mg

23

,34

80

.05

0.2

40

.30

59

0.4

27

0.6

21

0.8

91

.17

81

.36

51

.71

90

.69

0.3

40

.76

0.0

6

Mn

24

,04

94

16

32

01

27

63

84

53

16

97

78

59

67

41

7.4

11

88

.47

0.6

6-

0.0

9

Na

24

,25

30

.04

20

.33

0.4

26

0.6

0.8

21

.08

1.3

71

.54

1.8

59

0.8

60

.36

0.4

7-

0.1

8

Nb

12

,75

54

67

81

01

31

72

02

41

1.2

24

.13

0.9

20

.57

Ni

16

,23

12

68

11

15

20

26

30

40

15

.85

7.2

50

.83

0.6

0

Pb

20

,95

65

68

11

15

21

27

30

38

16

.58

7.3

70

.58

-0

.28

Sc

24

,25

60

.12

.63

45

.87

.39

.11

0.3

12

.95

.91

2.3

60

.49

-0

.10

Sr

13

,88

26

82

.15

10

31

34

18

52

77

38

94

48

54

42

15

.79

11

0.8

30

.94

0.2

2

Th

22

,58

71

.13

46

8.1

10

.71

31

4.9

19

.98

.38

3.6

20

.38

-0

.10

Ti

22

,91

91

91

45

51

,76

42

,24

32

,85

43

,67

64

,57

35

,12

46

,32

63

01

5.5

71

11

1.2

20

.52

0.1

5

U2

5,2

67

0.6

81

.71

22

.38

2.8

3.3

64

4.4

5.2

72

.90

0.7

90

.48

0.2

5

V2

3,2

62

22

63

24

15

36

78

49

51

16

55

.41

20

.37

0.5

80

.25

Zn

19

,91

83

26

32

43

58

75

96

11

01

35

61

.22

24

.82

0.6

20

.10

1492 Environ Geol (2009) 58:1479–1497

123

repository website of this paper (https://webspace.utexas.

edu/howarifm/www/NURE/1nm.htm/).

Spatial analysis

For Spatial analysis (Figs. 8, 9, 10, 11), the NURE data are

joined to the point coverage created by the GIS from lati-

tude and longitude data for each sample location.

Geochemical mapping of the data are produced using

ArcGIS. Maps showing the actual numerical concentration

of the elements at their locations are not produced in

regional geochemical mapping studies, rather, points or

sample locations are color-coded according to the con-

centration ranges of the element, with the highest range

shown in hot colors, and the lowest ranges shown in cold

colors. The geochemical data for stream sediments then are

interpolated in grid format to provide a graphical visuali-

zation of the regional variation in geochemical values. The

method of spatial interpolation is the inverse distance

weighted (IDW). The IDW techniques are recommended in

studies comparing interpolation methods (Robinson et al.

2004), and it available in the software. This method is

applied with 12 neighboring samples used for estimation of

each grid point. The power of one is chosen to acquire

some degree of smoothing effect. The color-scheme of

these maps is similar to that used in the point maps. These

maps are useful for defining regional trends, local anom-

alies, background, and providing a quick visual check of

the data processing.

Maps are also produced by color-coding polygons in a

pre-existing map. These maps are a geologic map, and a

drainage basin map base. Colors are assigned according

to the median concentration of the element in samples

falling within each polygon (Grossman 1998). Median

values are generally the best measure of central tendency

in regional geochemical datasets (Reimann and Filzmoser

2000). Median values of the chemical data are calculated

for map polygon areas using the Point-Stat-Calc exten-

sion for ArcView v.3.3 (Dombroski 2000). The legends

for the median value maps are based on the standard

deviation of the range in median values of data for

each of the polygons of grouped stream sediment sample

sites.

Spatial analysis here consists of several maps for each

chemical, which are (1) Point map of raw data; (2) IDW

grid map of the raw data; (3) Extraction of map 1 data for

each geologic polygon (Fig. 2) calculating descriptive

statistics for each polygon, and scaling the results; (4)

Extraction of map 1 data for each drainage basin (Fig. 3)

polygon, and scaling the results; (5) Scaled point map of

the outlier values. Spatial analyses maps for the examined

elements are presented at: https://webspace.utexas.edu/

howarifm/www/NURE/1nm.htm/ as well as Figs. 9, 10, 11.

Spatial distribution of factor scores

The five principal components of the data set from Table 4

are used in this section to show spatial distribution of the

factor scores. The factor scores for each sample are cal-

culated by summing the product of the loading or weights

shown in Table 5 multiplied by the corresponding geo-

chemical values. The factor scores for each sample are

plotted on maps at the data repository website of this paper,

and examples are being presented in the discussion section

of the present paper.

Median value rock chemistry for New Mexico

The simplified geologic map, Fig. 2, is layered over a map

of the chemistry of every element. The median value of

chemical concentration of all samples falling within a

polygon of a geologic unit is determined, and those values

are presented as the chemistry of that geologic unit.

Inherent questions are present, such as using the chemistry

of sediment to be equivalent to that of nearby bedrock.

Figure 12 consequently is for Mg, and Ce, and it shows the

median and range in concentrations of Mg and Ce in New

Table 4 Principal component loadings from PCA after rotation for

the maximum variance

PC1 PC2 PC3 PC4 PC5

Al 0.54 0.47 0.20 0.45 0.24

Ba 0.16 0.76 0.13 -0.04 0.13

Ca 0.17 -0.18 -0.70 0.04 -0.01

Ce 0.49 0.32 0.63 0.23 0.19

Co 0.86 0.22 0.12 0.05 0.16

Cr 0.87 -0.06 0.08 0.09 -0.14

Cu 0.69 0.10 0.00 0.36 0.18

Fe 0.84 0.36 0.06 -0.03 0.11

K -0.06 0.48 0.14 0.75 -0.07

La 0.44 0.32 0.68 0.24 0.15

Li 0.47 -0.04 -0.07 0.73 0.00

Mg 0.52 -0.13 -0.54 0.35 0.01

Mn 0.59 0.45 0.16 0.17 0.18

Na 0.07 0.82 0.11 0.18 -0.04

Ni 0.81 -0.08 -0.07 0.25 0.00

Pb 0.07 0.09 0.12 -0.01 0.93

Sc 0.83 0.06 0.24 0.22 0.06

Th 0.14 -0.19 0.46 0.30 0.27

Ti 0.62 0.53 0.28 -0.02 -0.17

U 0.26 0.07 0.73 -0.04 -0.01

V 0.84 0.26 0.07 -0.19 0.00

Zn 0.70 0.09 0.14 0.36 0.14

% of variance 33.25 12.58 12.38 9.64 5.68

Cumulative % 33.25 45.83 58.21 67.84 73.53

Environ Geol (2009) 58:1479–1497 1493

123

Mexico rock types. This diagram is presented for every

chemical element, and includes all the simplified rock

types. The Geologic Rock Units are derived from a mod-

ified Digital Generalized Geologic Map of United States,

Puerto Rico, and the US Virgin Islands, in ARC/INFO

Format (Reed and Bush 2005). The associated geologic

rock units categories are listed on the web: https://

webspace.utexas.edu/howarifm/www/NURE/1nm.htm/.

Discussion

Interpretation of the distribution of the chemical data has

mainly based on mineralogy, lithological units and geol-

ogy. The presented results indicate that the elements

reported here are present largely in minerals. However,

elements bounded to other geochemical phase of sediments

e.g. absorbed or colloidally bound chemicals may also be

present; adsorbed would increase with the iron content,

although this is probably minor in occurrence but believed

to exist. For example the outliers of Ce are concentrated in

the felsic igneous rock centers of the El Capitan area, and

the Organ Mountains. Outliers with moderate values are in

the Mogollon Volcanic Plateau. Another significant area of

outliers is in the Burro Mountains northeast of Lordsburg

which more prominent than the area of outliers is in the

Boot Heel of New Mexico. In the San Juan basin are many

outliers which are not easily explained. Young volcanic

rock of north central New Mexico also contains numerous

outliers. Whereas Mg is associated with Ca in carbonate

rocks and also it is rich in mafic igneous rocks. In southeast

New Mexico, Mg is associated with sedimentary rocks.

The dramatic number of outliers immediately north of

latitude 34� is influenced by the analytical methodology.

The outliers in the northwest corner of New Mexico are

associated with carbonate rocks of the San Juan basin.

Mineralogy consists of major minerals, quartz and clay,

minor minerals such as Fe or Mn oxyhydroxides, and

accessory, trace, or resistate minerals, such as magnetite,

zircon, rutile, and others. Multielement cluster analysis

separates the 22 chemical into 5 important groups (Fig. 8).

Although all the noted locations have outliers but they do

differ from one another mainly due to geological and

mineralogical conditions.

1. (Ce, La, Al, and Mn). This is a rare earth element

signature of peraluminous granites.

The REEs are found in resistate minerals in the sedi-

ment, such as monazite, zircon, and others.

2. (Fe, V, and Ti). Magnetite component present as a

resistate in the sediment.

3. (Co, Sc, Cr, Ni, Cu, and Zn). These chemicals are a

combined mafic geochemical indicator.

4. (K, and Li), (U, Th, and Pb). This is an alkali metal and

actinide signature of alkali granites.

5. (Na and Ba)

Fig. 12 The median and range concentrations of Mg and Ce in New

Mexico rock types

Table 5 Groups of elements based on the principal components

loading

5 factors incorporate of the 73.53% variance

Eigen

value

Major

elements

Trace elements Negative

F1 = 7.32 Al, Fe Co, Cr, Cu, Mg, Mn,

Ni, Sc, Ti, V, Zn

F2 = 2.77 Na Ba, Ti

F3 = 2.72 Ce, La, U Ca, Mg

F4 = 2.12 K Li

F5 = 1.25 Pb

1494 Environ Geol (2009) 58:1479–1497

123

6. (Ca and Mg). This is a carbonate signature. These two

elements belong to the same family in the column IIA

in the periodic table, and all are enriched in limestone.

They are mobile in the environment, and in sediments

they share similar behaviors of enrichment and

depletion.

Based on the Principal Component Analysis, the 22

elements can be grouped into five PCs. It was expected to

trace the elements back and forth between the cluster

analyses and PCA to large extent, but the results does not

support this; which indicate that not all the element control

the distribution or geochemical association in the studied

sediment in the same manner. The element groups pre-

sented in the PCA accounts for the major variances of

dataset as will be described next. The element classification

from PCA is reflecting part of the results from cluster

analysis. Table 5 shows the five groups of elements based

on PC loading.

1. Factor (1) (Al, Co, Cr, Cu, Fe, Mg, Mn, Ni, Sc, Ti, V,

and Zn)

2. Factor (2) (Ba, Na, and Ti)

3. Factor (3) (Ce, La and U), (Ca and Mg)

4. Factor (4) (K and Li)

5. Factor (5) (Pb)

Factor (1) accounts for 33.25% of the total variance,

and contains a high loading of Al, Co, Cr, Cu, Fe,

Mg, Mn, Ni, Sc, Ti, V, and Zn. This group can be

further subdivided into subfamilies, such as the

mafic trace elements with the presence of Co, Cr,

Fe, Sc, Ni, and V, ultramafic rocks with the presence

of Cr, Co, Ni, and Cu, (Levinson 1980; Thornton

1983).

Factor (2) accounts for 12.58% of the total variance,

and contains a high loading of Ba, Na, and Ti. These

elements have high concentrations in felsic igneous

rocks.

Factor (3) accounts for 12.38% of the total variance,

and contains a high loading of (Ce, La and U), and a

negative loading of (Ca and Mg). This group

suggests the rare earth element granite, are a family

in resistate minerals.

Factor (4) accounts for 9.64% of the total variance,

and contains a high loading of K and Li. These two

elements are the alkali metals characteristic of alkali

igneous rocks.

Factor (5) accounts for 5.68% of the total variance,

and contains a high loading of Pb with a value of

0.93. The fact that this lead component is singular,

and that it is not associated with Cu or Zn, denies the

possibility that it is associated with mineral deposits.

It is probably of anthropogenic origin, specifically

leaded gasoline. All gasoline contained lead in the

late 1970s. Factor 5 needs to be investigated more.

The five principal components of the data set are used in

this section to show spatial distribution of the factor scores.

The factor scores for each sample is calculated by summing

the product of the loading or weights shown in Table 3

multiplied by the corresponding geochemical values. The

first factor is dominated by Al, Co, Cr, Cu, Fe, Mg, Mn, Ni,

Sc, Ti, V, and Zn. Figure 13a shows high scores are closely

associated with the widespread, young, mafic igneous

rocks, with some variability, and indeed, this factor is a

mafic association of elements. The map shows very clear

and interesting regional structures and sharp boundaries

emerge. The most unusual feature is a clear liner zone

extending from the Raton volcanic field in the northeast

towards the Mogollon Volcanic Plateau, marking a linea-

ment which is known as the Jemez lineament. The Jemez

lineament is a northeast-trending zone characterized by

alignment of late Miocene through Quaternary bimodal

volcanic rocks (Pazzaglia and Hawley 2004). The third

factor is dominated by two groups which are Ce, La and U,

and negatively by Ca and Mg. Figure 13b shows the high

positive score with widespread felsic volcanic rock and no

limestone. This factor is dominated by the rare earth ele-

ments (REEs).

Future work

It has been said that research creates more questions than it

answers. The present research is a strong example of this,

and next are some additional questions to be considered by

the data.

1. Chemistry of the Great Plains. Sediments of the Great

Plains were deposited as a large alluvial fan east of the

rising Southern Rocky mountains. Later, the Pecos

River incised into this fan, and a divide is present east

of the Pecos, such that a cross section of the alluvial

fan is present to the west of that divide. Is there

variable chemistry in the fan?

2. From the above, can mineralogy be built from the

chemistry, because the mineralogy of the Great Plains

is relatively simple? Can remote sensing band combi-

nation maps which respond to mineralogy be merged

or layered with regional geochemical mapping?

3. Chemical footprint of large copper smelters. Two large

copper smelters are present in Southern New Mexico,

one over 100 years old, and the other young. Can those

footprint be revealed?

4. Special study area in the Grants Uranium District.

Higher sample density and better arsenic and uranium

analytical methodology is an opportunity for study.

Environ Geol (2009) 58:1479–1497 1495

123

Fig. 13 Spatial distribution of

high factor scores (the factor

scores for each sample are

calculated by summing the

product of the loading or

weights multiplied by the

corresponding geochemical

values; then factor scores for

each sample are plotted on

maps)

1496 Environ Geol (2009) 58:1479–1497

123

5. Correlation of geologic distribution of (young) mafics

and Mafic Factor.

6. Subsets having homogeneity of source rock and rock,

sediment, water chemistry provide unique geochemical

partition coefficients between sample types. One

example at Capitan has been accomplished. The rest

need to be investigated.

Conclusions

The interpretations of the regional geochemical mapping of

New Mexico from NURE datasets made possible by the

generation of various figures, maps, and tables. The study

found that PCA and GIS mapping of NURE stream sedi-

ment data are powerful tools in defining the regional and

local geochemical patterns related to the underlying geol-

ogy and anthropogenic sources. The study reported

geochemical anomalies that indicate several mineral belts

in New Mexico. These anomalies are clusters in several

areas including the Mogollon Volcanic Plateau, the Valles

caldera, the San Juan basin and the El Capitan Mountains,

as well as the main mining districts, and the major mining

industries present in the area. PCA shows a clear trend with

the association of the elements Co, Cr, Fe, Ni, Sc, and V.

This association is described as indicative of a mafic

chemistry signature. The mafic factor clusters in the Rio

Grande rift and Jemez lineament. On the other hand, the

REE factor consists of Ce, La and U, and it has strong,

localized clusters in the Organ Mountains, Boot Heel, San

Andres Mountains and El Capitan Mountains. The study

also concluded that the common REE elements are found

in certain felsic igneous rocks and in pegmatites. Mainly

the distribution of the elements in stream sediments in New

Mexico shows that most of the variability is controlled by

the bed rock chemistry.

References

Annual Report for the International Union of Geological Sciences

(IUGS) (2006) IUGS working group on global geochemical

baselines annual report 2006 8 December 2006; http://www.iugs.

org/PDF/Annual%20Report%202006%20GGB1.pdf

Bolivar SL (1980) An overview of the National Uranium Resource

Evaluation (NURE) hydrogeochemistry and stream sediment

reconnaissance for uranium in the United States. Los Alamos

Scientific Laboratories NURE Program, Report LA-8457 MS,

UC-51, pp 24

Bounessah M, Atkin BP (2003) An application of exploratory data

analysis (EDA) as a robust non-parametric technique for

geochemical mapping in a semi-arid climate. Appl Geochem

18:1185–1195

Cocker MD (1999) Geochemical mapping in Georgia, USA: a tool for

environmental studies, geologic mapping and mineral explora-

tion. J Geochem Explor 67:345–360

Dombroski M (2000) ESRI arc view extension: point Stat Calc. U.S.

geological survey open-file report 00-302

Grossman JN (1998) National geochemical atlas: the geochemical

landscape of the conterminous United States derived from

stream sediment and other solid sample media analyzed by the

National Uranium Resource Evaluation (NURE) program. U.S.

geological survey open file- report 98-622, version 3.01

Howarth RJ, Thornton I (1983) Regional geochemical mapping and

its application to environmental studies. In: Thornton I (ed)

Applied environmental geochemistry. Academic Press, New

York, pp 41–70

Kurzl H (1988) Exploratory data analysis: recent advances for the

interpretation of geochemical data. J Geochem Explor 30:309–

322

LASL (1977) Estancia valley pilot survey, New Mexico. NURE/

HSSR Program, GJBX-21(77)

LASL (1981a) Grants special study, New Mexico. NURE/HSSR

Program, GJBX-351(81)

LASL (1981b) San Andres and Oscura Mountains detailed survey,

New Mexico. NURE/HSSR program, GJBX-215(81)

Levinson AA (1980) Introduction to exploration geochemistry, 2nd

edn. Applied Publishing Ltd, Illinois, pp 615–924

Ludington S, Folger H, Kotlyar BG, Mossotti V, Coombs MJ,

Hildenbrand TG (2006) Regional surficial geochemistry of the

Northern Great Basin. Econ Geol 101:33–57

Mack GH, Giles KA (2004) Geology of New Mexico, A geologic

history, New Mexico Geological Society Special Publication,

fiftieth anniversary volume. New Mexico Geological Society

Special Publication

Pazzaglia FJ, Hawley JW (2004) Neogene and quaternary geology

and geomorphology: the geology of New Mexico, New Mexico

Geological Society, 407–443

Reed JC Jr, Bush CA (2005) Generalized geologic map of the

conterminous United States: U.S. Geological Survey, Denver,

CO. \http://pubs.usgs.gov/atlas/geologic/[Reimann C, Filzmoser P (2000) Normal and lognormal data

distribution in geochemistry: death of a myth. Consequences

for the statistical treatment of geochemical and environmental

data. Environ Geol 39:1001–1014

Reimann C, Filzmoser P, Garrett RG (2004) Background and

threshold: critical comparison of methods of determination. Sci

Total Environ 346(1–3):1–16

Ried JC (1993) A geochemical atlas of North Carolina, USA.

J Geochem Explor 47:11–27

Robinson GR Jr, Kapo KE, Grossman JN (2004) Chemistry of stream

sediments and surface waters in New England, U.S. geological

survey, open-file report 2004-1026. \http://pubs.usgs.gov/of/

2004/1026/[Sall J, Lehman A (1996) A guide to statistic and data analysis using

JMP� and JMP in software SAS Institute Inc. Duxbury Press, CA

Shacklette HT, Boerngen JG (1984) Element concentration in soils

and other surficial materials of the conterminous United States.

U.S. geological survey, professional paper 1270, 35 pp

Thornton I (1983) Applied environmental geochemistry. Academic

Press, New York, pp 51–133

Tukey JW (1977) Exploratory data analysis. Addison-Wesley, Reading

USGS (2004) The national geochemical survey-database and docu-

mentation. U.S. geological survey open-file report 2004-1001

Xie X, Ren T (1993) National geochemical mapping and environ-

mental geochemistry—progress in China. J Geochem Explor

49(1–2):15–34

Environ Geol (2009) 58:1479–1497 1497

123